Trying to diagnose horizontal line artifacts in VHS capture

[alastair]

I'm newly registered, and I didn't seem to find any existing threads that were immediately relevant. So if this posting should belong to an existing thread, please let me know.

I'm trying to figure what the failure is in the recording of the attached captured video samples, to work out how to correct it. Probably in ffmpeg, which I'll be using anyway to trim the recordings and re-encode them down to reasonable size.

After a lot of difficulty with the Diamond I finally got a (mostly) working arrangement under Ubuntu (which of course Diamond themselves don't support ). With vlc and video4linux2, and the refinement of a very exact series of steps to do it, I've been able to record 2 or 3 VHS tapes to 720x480 format, via component cables.

(NOTE that the Diamond VC500, while showing its blue power light very brightly -- that seems to be common to every mention of it I've seen -- doesn't seem to react to the RECORD button at all: the 2nd LED never lights.)

But they all have an unexpected video artifact: a set of horizontal lines that seem to cross the entire frame, apparently smearing video content to the left, and at different vertical positions in each frame; as demonstrated by the 3 sequential frames I've posted here, captured by VLC:

[Attachment 33177 - Click to enlarge]

[Attachment 33178 - Click to enlarge]

[Attachment 33179 - Click to enlarge]
As far as I can detect, there is no corresponding audio anomaly. The audio is far from wonderful, but I expect that's in the original tape.

It's not in the VHS player or the component cables that connect it to the VC500: I've connected it and them to our Aquos flat screen TV, and the tapes I try (including this one of course) play without perceptible artifacts.

I've tried doing the capture with the VC500 in Windows, but none of the software I've tried (vlc, EZ Grabber, ArcSoft ShowBiz, VirtualDub, etc) even winds up displaying a single image from the player. I have an open maintenance ticket with Diamond about these right now, but so far, little headway. I can give more details about this if wanted.

I'm not too worried about that part, though, as long as I can have it working smoothly in Ubuntu, which overall behaves better than Windows. Since Ubuntu can mount and write to Windows NTFS filesystems, the only inconvenience is that others using this machine need it rebooted back to Windows.

I can give more detail if it helps, but I think first I should be sure I'm posting in the right place.

Thanks in advance for all help,

Alastair

[pandy]

Various reasons - mechanical and electrical can be source for this problems... audio signal is stored usually in a way where even this kind of artifacts will be not perceived.
I would start from cleaning and eventually adjusting VCR (so proper service/maintenance).

[jagabo]

Most capture devices are prone to problems like this with small time base errors. You need a line time base corrector. Either one built into an old S-VHS deck or an old DVD recorder (in passthrough mode) like the Panasonic ES10 or ES15.

[friendly_jacek]

I've been able to record 2 or 3 VHS tapes to 720x480 format, via component cables.

i'm impressed, i didn't know VCR supported component cables. is this a D-VHS?

[alastair]

I've been able to record 2 or 3 VHS tapes to 720x480 format, via component cables.

i'm impressed, i didn't know VCR supported component cables. is this a D-VHS?

Firstly, sorry for the long delay. I had been starting to work on this when I found a filesystem consistency problem with Windows, and had my antivirus crashing on startup. It's taken this long to ferret out what I hope is the solution.

Anyway, back to business:

Whoops, I just realised what "component" cables sounds like: by "component" cables I didn't mean RGB, I meant the yellow-white-red triple for video-left-right components. As opposed to S-Video. No, this is an analogue Philips VCR: no RGB components possible.

Alastair

[alastair]

Most capture devices are prone to problems like this with small time base errors. You need a line time base corrector. Either one built into an old S-VHS deck or an old DVD recorder (in passthrough mode) like the Panasonic ES10 or ES15.

Hmm, hmm, and again hmm. What I'm using right now is a Philips that I bought new just for this purpose. We've had others in the past but by now they're all defunct. (Except a rather old one integrated into a TV, which I didn't want to trust).

Now I did take the Philips into the family room to add it into the TV's setup (nice big Sharp Aquos display) and there the display was clean: no horizontal distortions.

However, I don't have a means of getting the desktop (tower, actually) that I'm using to record into a connection with that system. I don't believe it even provides an output.

Attempted Solution:
What I have done is bought a $30-40 Digital Video Stabiliser from Amazon that got a very large number of high reviews. It does everything automatically: there aren't even controls to use. You just plug the video feed from the VHS into its IN, and the video supply to the Diamond CV500 to its OUT. The one qualification is that it claims to be specifically for VHS to DVD connection; however I presume a VHS to Diamond CV500 is the same functionality.

The result?

No Change. The crew of dancing horizontal lines is still there. No visible change at all.

Now, there's one thing I don't think I mentioned before:
These lines appear without the tape playing, when the player is just sitting in blue screen where the STOP and PLAY keywords display.

Have I got a bad CV500?

I've looked a few times at the Legato, but it's so expensive, at least for me.

Thanks again for the help. I have to make this work.

Alastair

[alastair]

Various reasons - mechanical and electrical can be source for this problems... audio signal is stored usually in a way where even this kind of artifacts will be not perceived.
I would start from cleaning and eventually adjusting VCR (so proper service/maintenance).

Even though it was newly bought for this project?

Also, as I mentioned earlier, I connected this same VCR to our widescreen LCD TV and played the same tape in it, and the artifact was gone.

Thanks,
Alastair

[PuzZLeR]

You can indeed try a different device that isn't expensive, instead of your VC500, such as the Hauppauge USB-Live2, ezcap.tv (the REAL one from its name's site), StarTech SVID2USB2 or ATI USB 600*. All are among the better ones at capturing VHS video. I personally won't bet it's the solution to your problem, but at worst could likely be better quality than the VC500 in the end.

Or, you can see if there's some electrical interference around where your capture setup is - even running a house fan nearby can cause problems.

However, I do think your solution was pointed out by Jagabo in post 3. VHS video may run fine under conventional playback in its humble analog world, but becomes a different, and very chaotic, beast when there are attempts to digitize it.

A digital video stabilizer, such as the one you bought, can be useful in that it works at defeating MacroVision and false positives, but is not synonymous with a TBC at the line level - which is likely what is needed to solve your problem.

*No longer in production, and can only be bought used or through resellers of back stock, and may only run well on older systems, but still a very good device.

[jagabo]

Try this easy fix first: Fast forward the tape to the end. Rewind the tape back to the beginning. That will loosen the tape in the case and may be enough to make it play more smoothly.

Try capturing the same problematic section two or three times. Examine the captures. Do the time base errors appear in exactly the same frame each time? If not, you can use three captures and a median filter to remove most of them.

[PuzZLeR]

Try capturing the same problematic section two or three times. Examine the captures. Do the time base errors appear in exactly the same frame each time? If not, you can use three captures and a median filter to remove most of them.

Yes indeed. If it's just "lines" or "dots", or other similar random errors, in different places each time you run the tape, then yes, a median filter is the ultimate solution and will surely correct this. Or, better yet, do five captures if the tape is in particularly bad shape.

Here's a good start:
https://forum.videohelp.com/threads/362361-Median%28%29-plugin-for-AviSynth

@alastair: If you'd like to upload three separate small captures of the same scene, I, or even someone else here, can give it a go for you. (Even if I can "fix" it, I have a feeling it will still need line based correction.)

[jagabo]

Even if I can "fix" it, I have a feeling it will still need line based correction.

Yes, of course. It's VHS.

[PuzZLeR]

Even if I can "fix" it, I have a feeling it will still need line based correction.

Yes, of course. It's VHS.

Oh, I've actually made line based correction a default to all VHS I'd capture now, even if the tape "plays fine". But I have a feeling that the problem here will more-so need line based correction, beyond the regular flaws inherent in VHS, even if median methods do help.

Would love to have some samples.

[alastair]

Try this easy fix first: Fast forward the tape to the end. Rewind the tape back to the beginning. That will loosen the tape in the case and may be enough to make it play more smoothly.

The motion of the tape itself doesn't seem impaired. It has been back and forth a few times now,
as I keep retrying to diagnose it. And, as I may have mentioned, I took that tape and that VHS player,
the Philips, next door to play it on our widescreen TV. Nice display, no artifacts.

Try capturing the same problematic section two or three times. Examine the captures. Do the time base errors appear in exactly the same frame each time? If not, you can use three captures and a median filter to remove most of them.

Now that's an interesting thought, if I'm following you rightly. Are you thinking that in different passes a given frame might suffer
artifacts in different places? (If you're saying that successive frames even in the same pass may have them in different places,
then I know they do -- but I assume a median filter would only work if given the same intended content every time.)

I'll need to look at that, though I'll need to work out how to catch exactly the same frames in different passes.

The first obvious problem, while not actually a functional one, is that this would more than triple the time necessary
to record each tape. Because the problematic section is in fact everything from when the tape goes in and the PLAY
screen appears to before the tape is ejected. Still, worth looking at.

[jagabo]

Try this easy fix first: Fast forward the tape to the end. Rewind the tape back to the beginning. That will loosen the tape in the case and may be enough to make it play more smoothly.

The motion of the tape itself doesn't seem impaired. It has been back and forth a few times now,
as I keep retrying to diagnose it.

You need to go all the way to the end then back to the start in one go. If you've been capturing short sections then rewinding and capturing again, or capturing different section here and there by FF and RW a little, you are making the problem of uneven winding of the tape worse, not better. This may not help but the point is that this is an easy thing to try.

And, as I may have mentioned, I took that tape and that VHS player,
the Philips, next door to play it on our widescreen TV. Nice display, no artifacts.

TVs are much more forgiving of small time base errors that most capture devices. The horizontal time base errors in your samples are on the order of tens of nanoseconds. That's too small to be audible, but easily visible in a capture if uncorrected. The phase locked loop in the TV can easily adjust for that.

Try capturing the same problematic section two or three times. Examine the captures. Do the time base errors appear in exactly the same frame each time? If not, you can use three captures and a median filter to remove most of them.

Now that's an interesting thought, if I'm following you rightly. Are you thinking that in different passes a given frame might suffer
artifacts in different places?

Yes. If the artifacts are in different places each time and there aren't too many of them, then only one of the three recordings is likely to show any particular artifact at any particular location. Picking the median of three then gives you one of the good videos. This is done on a pixel-by-pixel basis -- ie, each pixel is compared in three frames and the median is selected.

I'll need to look at that, though I'll need to work out how to catch exactly the same frames in different passes.

The easiest way to do that is to open the caps in multiple instances of an editor (VirtualDub is good for this) side-by-side then step through the videos frame-by-frame (left and right arrow keys in VirtualDub) and compare. The caps don't have to start at exactly the same frame for this -- since you can align them manually.

The first obvious problem, while not actually a functional one, is that this would more than triple the time necessary
to record each tape.

Yes. You have to decide if you want a clean(er) video or a quick cap.

[PuzZLeR]

Jagabo saved me a bit of typing - thanks.

And Jagabo is bang on, so I won't be redundant, only add to this:

The first obvious problem, while not actually a functional one, is that this would more than triple the time necessary to record each tape. Because the problematic section is in fact everything from when the tape goes in and the PLAY screen appears to before the tape is ejected. Still, worth looking at.

Definitely worth looking at. I always capture each tape at least 3X. If more problematic, or more important, 5X.

There is a HUGE difference in quality. And even if it doesn't totally correct this problem, it still always corrects many random problems.

It can cure dropped frames during captures (I do this manually), random dropouts, random snow, random black streaks, tracking bursts, much of the chroma dancing, shimmer noise, and much more. I once had an electrical interference ruin a capture once with white spots when I wasn't looking, and still didn't re-capture it because the median method cleaned it up. That's confidence!

As for the work, or more than "necessary", it's no big deal. The only real work is setting up the first capture, with your settings, etc., which you'd do anyway. Once that was successful, the second, third, or any other after is as easy as rewinding and replaying and walking away for some time and coming back later. I would even grocery shop or something else during these subsequent captures.

To me, it's very finite "work", not forever - only till one finishes their VHS tapes and moves on. In the long run you'll be glad you did it, and won't care.

If you like, capture a small scene three times, post the caps, and we'll do our best to help you set it up.

[alastair]

Even if I can "fix" it, I have a feeling it will still need line based correction.

Yes, of course. It's VHS.

Oh, I've actually made line based correction a default to all VHS I'd capture now, even if the tape "plays fine". But I have a feeling that the problem here will more-so need line based correction, beyond the regular flaws inherent in VHS, even if median methods do help.

Would love to have some samples.

"Line base correction" -- is that the same as "time base correction"? If so, how do you achieve it for every new digitisation? I tried to get some economical hardware for it from Amazon, but it turned out I didn't read far enough to see the user q&a where it was specifically reported NOT to do TBC. Waste of money. I went to look for another one, a little less economical, but the only one I saw in reasonable range is apparently using the term TB correction in a very loose fashion. Not buying it.

Do you have software that can do the relevant TB correction? Thanks.

[alastair]

Jagabo saved me a bit of typing - thanks.

And Jagabo is bang on, so I won't be redundant, only add to this:

The first obvious problem, while not actually a functional one, is that this would more than triple the time necessary to record each tape. Because the problematic section is in fact everything from when the tape goes in and the PLAY screen appears to before the tape is ejected. Still, worth looking at.

Definitely worth looking at. I always capture each tape at least 3X. If more problematic, or more important, 5X.

There is a HUGE difference in quality. And even if it doesn't totally correct this problem, it still always corrects many random problems.

It can cure dropped frames during captures (I do this manually), random dropouts, random snow, random black streaks, tracking bursts, much of the chroma dancing, shimmer noise, and much more. I once had an electrical interference ruin a capture once with white spots when I wasn't looking, and still didn't re-capture it because the median method cleaned it up. That's confidence!

Sounds very hopeful. Who'd have expected randomness to be an advantage ?

I was going to try doing this, but my Diamond VC500 cxt is suddenly playing silly buggers and giving me loads of snow, bad colour ranges, absurd brightness, tearing,
etc. I've been trying (through VLC) to work out the necessary parameters, but that's a lot of variables to juggle. I actually got a good picture last night, for the first time in a while, but I can't be sure exactly what caused it.

I actually tried bringing in another, older VCR (whose tape carriage broke), but which is generally a lot more sophisticated than my current one, with the idea that I'd try piping the Philip's signal through it and thence to the Diamond, hoping it would turn out to have the TBC circuitry, and would just pass the signal through if I pressed REC. (It only later occurred to me that the last thing I want is 2 levels of analogue signals in a row). That's when the Diamond apparently started acting up. I gave it up and removed the older VCR from the system.

As for the work, or more than "necessary", it's no big deal. The only real work is setting up the first capture, with your settings, etc., which you'd do anyway. Once that was successful, the second, third, or any other after is as easy as rewinding and replaying and walking away for some time and coming back later. I would even grocery shop or something else during these subsequent captures.

Oh indeed, the "work" wasn't really on my mind. It's the number of tapes we have, and what's probably going to be a finite amount of time to digitise them. Along with scanning paper documents, digitising old audio cassettes, and digitising our LP collection (I'm actually about halfway through this one, though still lots of cleaning to do in the digital versions).

To me, it's very finite "work", not forever - only till one finishes their VHS tapes and moves on. In the long run you'll be glad you did it, and won't care.

As long as it all works I expect that'll be true. I'll have rather a pile of DVDs to burn but that should be OK. (I don't want to leave them on the hard drive of any one machine). I'm looking forward to having them safe from VHS degradation over time.

If you like, capture a small scene three times, post the caps, and we'll do our best to help you set it up.

Thank you very much, I certainly would like. Now if I can just get the bleeping Diamond capture back to working as it did when I sent those 3 samples.

Which leads me to a related pair of questions, hoping nobody will mind me adding them in on this same thread:

1. Is anybody else having this kind of trouble with the Diamond VC500 or does it normally work acceptably?
2. I'm thinking seriously of replacing the Diamond with the Elgato. A lot more expensive but apparently a better reputation. Anybody have any reactions about that, good or bad?

Thanks again for your time and expert advice.

Alastair

[jagabo]

"Line base correction" -- is that the same as "time base correction"?

line vs. full frame tbc:
https://forum.videohelp.com/threads/338608-What-would-you-do-with-this-captured-video-%...=1#post2104575
https://forum.videohelp.com/threads/306272-Computer-video-capture-vs-vcr-to-dvd-combo?p...=1#post1882662
https://forum.videohelp.com/threads/319420-Who-uses-a-DVD-recorder-as-a-line-TBC-and-wh...=1#post1983288

[alastair]

"Line base correction" -- is that the same as "time base correction"?

line vs. full frame tbc:
https://forum.videohelp.com/threads/338608-What-would-you-do-with-this-captured-video-%...=1#post2104575
https://forum.videohelp.com/threads/306272-Computer-video-capture-vs-vcr-to-dvd-combo?p...=1#post1882662
https://forum.videohelp.com/threads/319420-Who-uses-a-DVD-recorder-as-a-line-TBC-and-wh...=1#post1983288

Thank you, very interesting. I'll need to study them more.